← 返回 Skills 市场

Multimodal Asset Tagger

Name: Multimodal Asset Tagger
Author: geoly-geo

作者 GEOLY AI · GitHub ↗ · v1.0.0

cross-platform ⚠ suspicious

407

总下载

当前安装

版本数

在 OpenClaw 中安装

/install geo-multimodal-tagger

功能描述

Generate AI-optimized Alt Text, file names, captions, and Schema markup for images, videos, and audio assets. Improves AI discoverability on Google Lens, Cha...

使用说明 (SKILL.md)

Multimodal Asset Tagger

Methodology by GEOly AI (geoly.ai) — every image and video is a citation opportunity AI can either read or miss.

Generate optimized metadata for images, videos, and audio files for AI platforms.

Quick Start

python scripts/optimize_asset.py --type image --description "dashboard showing metrics" --output optimized.md

Why Multimodal Matters

AI platforms increasingly read visual content:

Platform	Visual Capability	Citation Type
Google Lens	Image search	Direct image citation
ChatGPT Vision	Image understanding	Contextual reference
Perplexity	Video transcripts	Transcript citations
Gemini	Native image processing	Multimodal answers

Image Optimization

Alt Text Formula

[Descriptive subject] + [Brand if relevant] + [Context/use case]

Examples:

❌ alt="image1.jpg"
❌ alt="product photo"
✅ alt="GEOly AI dashboard showing AIGVR score trend over 30 days"
✅ alt="Brand visibility comparison chart across ChatGPT and Perplexity — GEOly AI"

Filename Formula

[primary-keyword]-[secondary-keyword]-[brand]-[descriptor].jpg

Examples:

❌ IMG_3847.jpg
✅ geo-brand-visibility-dashboard-geoly-ai.png
✅ aigvr-score-chart-ai-search-monitoring.jpg

ImageObject Schema

{
  "@context": "https://schema.org",
  "@type": "ImageObject",
  "name": "AIGVR Score Dashboard",
  "description": "Dashboard showing brand visibility scores across AI platforms",
  "contentUrl": "https://example.com/images/dashboard.jpg",
  "author": {
    "@type": "Organization",
    "name": "GEOly AI"
  },
  "keywords": "AIGVR, brand visibility, AI search, dashboard"
}

Video Optimization

Checklist

Title contains primary keyword
Description: first 150 chars = keyword + brand
Transcript/captions attached (SRT/VTT)
Chapters/timestamps for long videos
Thumbnail: keyword-rich filename
VideoObject Schema added

VideoObject Schema

{
  "@context": "https://schema.org",
  "@type": "VideoObject",
  "name": "How to Optimize for AI Search",
  "description": "Complete guide to GEO strategies...",
  "thumbnailUrl": "https://example.com/thumbs/geo-guide.jpg",
  "uploadDate": "2024-01-15",
  "duration": "PT12M30S",
  "contentUrl": "https://example.com/videos/geo-guide.mp4"
}

Audio/Podcast Optimization

Descriptive episode titles (not "Episode 47")
150+ word descriptions, keyword-rich
Full transcript as page content
Guest names and topics as entities

Asset Optimization Tool

python scripts/optimize_asset.py \
  --type [image|video|audio] \
  --description "Asset description" \
  --brand "BrandName" \
  --keywords "keyword1,keyword2"

Output:

Optimized Alt Text
Recommended filename
Schema markup
Discoverability score (Before/After)

Scoring

Factor	Weight	Best Practice
Descriptiveness	30%	Specific, detailed
Keyword presence	25%	Natural inclusion
Brand mention	20%	When relevant
Context	15%	Use case clear
Length	10%	100-150 chars for Alt

Discoverability Score: 0-10

8-10: Excellent
6-7: Good
4-5: Fair
\x3C4: Poor

安全使用建议

This skill appears harmless and consistent with generating simple alt text/filenames, but the documentation over-promises features the code doesn't implement (schema markup, scoring, video/audio support). Before installing: (1) verify the source or owner if you require provenance; (2) review and run the small Python script in a sandbox/local environment—it only prints an alt text and a filename; (3) if you need Schema markup, scoring, or video/audio handling, plan to extend the code or use a different tool; (4) avoid supplying secrets or connecting it to production systems until you expand and test its functionality.

功能分析

Type: OpenClaw Skill Name: geo-multimodal-tagger Version: 1.0.0 The `scripts/optimize_asset.py` file generates filenames from user-provided descriptions without adequate sanitization. While the script itself only prints to stdout, this creates a path traversal or command injection vulnerability if the OpenClaw agent or a downstream system uses the generated filename directly to create files or execute shell commands. The `SKILL.md` instructs the agent to execute this script with user-controlled input, making this a potential attack vector. There is no evidence of intentional malicious behavior like data exfiltration or persistence.

能力评估

ℹ Purpose & Capability

The name and description (generate alt text, filenames, captions, and Schema markup for images, video, audio) align with the provided SKILL.md guidance. However, the only executable code (scripts/optimize_asset.py) only produces simple Alt Text and a filename for images; it does not produce Schema markup, discoverability scores, or explicit support for video/audio. This is an over-promise vs. actual capability.

ℹ Instruction Scope

SKILL.md instructs running the bundled Python script and contains methodology and templates. The runtime instructions do not request any secrets, system files, or network endpoints. They do, however, instruct generation of outputs (Schema, scores) that are not produced by the script, so following the SKILL.md may give a false expectation of behavior.

✓ Install Mechanism

No install spec is provided (instruction-only). The included Python script is small, pure local code, and there are no external downloads, package installs, or archive extraction steps.

✓ Credentials

The skill requests no environment variables, no credentials, and references no config paths. The code does not access environment variables or external services, so requested privileges are minimal and proportionate.

✓ Persistence & Privilege

The skill is not always-enabled and does not request persistent presence or modify agent/system configuration. It runs a local script when invoked and does not store credentials or alter other skills.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install geo-multimodal-tagger
安装完成后，直接呼叫该 Skill 的名称或使用 /geo-multimodal-tagger 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release: Generate Alt Text, filenames and Schema for images/videos, GEOly AI Multimodal Agent approach

元数据

Slug geo-multimodal-tagger

版本 1.0.0

许可证 —

累计安装 0

当前安装数 0

历史版本数 1

常见问题